Prompt Engineering
Lesson 1How to Get Exceptional Work Out of an AI Assistant
By the end you will have a repeatable method for turning any task into great AI output.
See it first
Most people use AI like a search box: type a one-liner, take whatever comes back, feel underwhelmed. The people who get exceptional work treat it like a thinking partner: they bring it into a process, step by step, the way they would brief a sharp junior colleague.
The model did not change between those two people. The process did. Prompt engineering is mostly about running a better process, and that process is learnable.
The mindset shift
A search box answers a query. A thinking partner helps you do work. That means: give it context, let it propose, react to its proposals, and steer toward the result, rather than expecting a perfect answer from a cold one-liner.
The methodology: break the work into steps
Exceptional output rarely comes from one prompt. It comes from a short sequence. Use this five-step method for anything that matters:
- Idea generation. Ask for many options, angles, or approaches before committing. "Give me 10 ways I could frame this."
- Refinement. Narrow and sharpen. "Combine 3 and 7, and make it bolder."
- Draft. Now ask for the full thing, with your chosen direction and context.
- Iterate. Steer the draft: shorter, warmer, restructure, add a section.
- Produce. Finalize the format and polish for the real audience.
Search-box prompt
"Write a strategy memo about expanding to Europe."
→ Generic, hollow, ignores your actual situation.
Methodology
Ideate angles → pick two → draft with your real context → iterate tone and structure → produce in memo format.
→ Specific, sharp, yours.
What you can now do
- Explain the difference between using AI as a search box and as a thinking partner
- Apply the five-step methodology: idea, refine, draft, iterate, produce
- Break a big task into a short sequence of prompts
- Recognize why one-shot prompting underperforms a guided process
Check your understanding
1 / 2What is the core mindset shift this lesson teaches?
What's next
You have a method for one task. Next, scale it across multiple steps that feed each other: Decomposition & Prompt Chaining.